Download Simulated Interactions between Angiotensin

Biochemistry 2007, 46, 8753-8765
8753
Simulated Interactions between Angiotensin-Converting Enzyme and Substrate
Gonadotropin-Releasing Hormone: Novel Insights into Domain Selectivity†
Athanasios Papakyriakou,‡ Georgios A. Spyroulias,*,‡ Edward D. Sturrock,| Evy Manessi-Zoupa,§ and
Paul Cordopatis‡
Departments of Pharmacy and Chemistry, UniVersity of Patras, Panepistimioupoli-Rion, GR-26504, Greece, and
DiVision of Medical Biochemistry, Institute of Infectious Disease and Molecular Medicine, UniVersity of Cape Town,
Rondebosch 7701, South Africa
ReceiVed February 6, 2007; ReVised Manuscript ReceiVed May 2, 2007
ABSTRACT:
Human angiotensin-I converting enzyme (ACE) is a central component of the renin-angiotensin
system and a major target for cardiovascular therapies. The somatic form of the enzyme (sACE) comprises
two homologous metallopeptidase domains (N and C), each bearing a zinc active site with similar but
distinct substrate and inhibitor specificities. On the basis of the recently determined crystal structures of
both ACE domains, we have studied their complexes with gonadotropin-releasing hormone (GnRH), which
is cleaved releasing both the protected NH2- and COOH-terminal tripeptides. This is the first molecular
modeling study of an ACE-peptide substrate complex that examines the structural basis of ACE’s
endopeptidase activity and offers novel insights into subsites that are distant from the obligatory binding
site and were not identified in the crystal structures. Our data indicate that a bridging interaction between
Arg500 of the N-domain and Arg8 of GnRH that involves a buried chloride ion may account for its role
in the specificity of the N-domain for endoproteolytic cleavage of the substrate at the NH2-terminus in
Vitro. In support of this, the protected NH2-terminal dipeptide of GnRH exhibits stronger interactions
than the protected COOH-terminal dipeptide with the N-domain of ACE. Further comparison of the models
of ACE-substrate complexes promotes our understanding of how the two domains differ in their function
and specificity and provides an extension of the pharmacophore model used for structure-based drug
design up to the S7 subsite of the enzyme.
Hypertension is a major risk factor in cardiovascular and
kidney disease, affecting a quarter of the world’s adult
population (1). The development of novel therapeutic approaches has focused on the renin-angiotensin system
(RAS), which plays a key role in the regulation of blood
pressure and electrolyte balance in humans (2). In the main
pathway of the RAS, renin acts on the circulating precursor
angiotensinogen to generate the decapeptide angiotensin
(Ang) I, which is converted by angiotensin-I converting
enzyme (ACE1) to the potent vasopressor octapeptide Ang
II. In parallel, ACE affects blood pressure by inactivating
bradykinin (BK) and kallidin, vasodilator peptides which are
generated from kininogen by the action of kallikrein (3).
Because of this dual function, ACE inhibitors have been first
line antihypertensive therapies for many years (4-7).
ACE (EC 3.4.15.1) is classified as a member of the M2
gluzincin family within the MA clan (8, 9). There are two
† A.P. has been supported by a postdoctoral scholarship from the
Greek National Foundation of Scholarships (IKY).
* To whom correspondence should be addressed. Tel: +30
2610969950and-951.Fax: +302610969950.E-mail: G.A.Spyroulias@upatras.
gr.
‡ Department of Pharmacy, University of Patras.
§ Department of Chemistry, University of Patras.
| University of Cape Town.
1
Abbreviations: ACE, angiotensin-I converting enzyme; sACE,
somatic ACE; tACE, testis ACE; nACE: N-domain of sACE; cACE:
C-domain of sACE; GnRH: gonadotropin-releasing hormone; Ang:
angiotensin, BK: bradykinin; rmsd: root-mean-square deviation.
isoforms of ACE that are transcribed from the same gene in
a tissue-specific manner (10): the somatic form (sACE) that
is found in a variety of tissues and the testicular form (tACE),
which is exclusively expressed in germinal cells. sACE is a
1,277-residue type I transmembrane glycoprotein with an
ectodomain consisting of two highly homologous parts (N
and C domains). tACE is a 701-amino-acid isoform that
except for the first 36 residues is identical to the C domain
of sACE (11). Each domain contains an active site bearing
the characteristic HEXXH zinc-binding motif of Zn-peptidases (zincins) (12). The sequence and structural homology
at the active sites of ACE favors an enzymatic mechanism
analogous to that proposed for thermolysin (13). This
hydrolytic reaction proceeds via a general-base mechanism
with the nucleophilic attack of a water molecule or hydroxide
ion on the carbonyl carbon of the scissile bond (14).
The recent breakthrough in determining the high-resolution
crystal structures of human tACE (C domain of sACE) (15)
and that of the N domain of sACE (16), both in the absence
and presence of the potent inhibitor lisinopril, has renewed
the interest in the study of their enzymatic activity at a
molecular level and has provided a structural basis for the
design of domain-specific inhibitors (17, 18). Despite the
structural homology of the two domains of sACE, which
have ∼60% sequence identity, some notable differences
between the active sites were observed (16). Consistent with
the observed chloride-dependent activation of ACE (19), only
10.1021/bi700253q CCC: $37.00 © 2007 American Chemical Society
Published on Web 07/03/2007
8754 Biochemistry, Vol. 46, No. 30, 2007
Papakyriakou et al.
Table 1: Sequences of the Natural Substrates Ang I, Ang1-7, BK,
and GnRH in Two Orientations for C-Terminal (GnRHC) and
N-Terminal Endopeptidase Cleavage (GnRHN)a
P8
P7
Asp Arg
P6
P5
Val Tyr
Asp
Arg Pro Pro
pGlu His Trp
nGly Pro Arg
P4
Ile
Arg
Gly
Ser
Leu
P3
His
Val
Phe
Tyr
Gly
P2
P1
P1′
Pro
Tyr
Ser
Gly
Tyr
Phe8
His9
Ile5
Pro7
Leu7
Ser4
His6
Phe8
Arg8
Trp3
P2′
P3′
Leu
Ang I
Pro
Ang 1-7
Arg
BK
Pro nGly GnRHC
His pGlu GnRHN
a
pGlu is the N-terminus pyroglutamic acid, and nGly is the amidated
C-terminus Gly residue of GnRH.
one chloride ion was bound to the N domain as opposed to
the two found in the crystal structure of tACE (16). In
addition, tACE has been co-crystallized with the antihypertensive drugs captopril and enalaprilat (20).
The most common small-molecule inhibitors exhibit
comparable selectivity for both active sites (21), whereas the
N domain is 1000-times more sensitive to inhibition by the
phosphinic peptide RXP407 (22) and more than 3000-times
less sensitive to inhibition by RXPA380 (23). Both domains
hydrolyze the principal physiological peptides Ang I and BK
at comparable rates but with different chloride concentration
requirements (24). Whereas the enzymatic activity of the C
domain is highly dependent on chloride concentration, that
of the N domain is much less so (24).
Interestingly, ACE exhibits endopeptidase activity for
substrates with amidated C-termini such as substance P and
gonadotropin-releasing hormone (GnRH, formerly known as
luteinizing hormone-releasing hormone or LHRH) by cleaving their C-terminal tripeptides. Other natural substrates with
high specificity for the N domain are Ang1-7 (25) and
N-acetyl-SDKP, the latter being involved in the control of
the hematopoietic stem cell proliferation (26). Because of
its protected amino-terminal pyroglutamic acid, GnRH is also
metabolized to release the N-terminal tripeptide, in addition
to the C-terminal tripeptide (Table 1) (27). The N domain
of ACE is mainly responsible for the primary amino-terminal
endoproteolytic cleavage of GnRH, whereas carboxyterminal cleavage is performed by both active sites of ACE
in Vitro (24, 28).
The unique endopeptidase activity of ACE for GnRH, in
conjunction with the available crystal structures of both
enzymatic domains, has motivated our in silico studies of
their complexes. With the aim of gaining structural information regarding the interaction of the enzyme-substrate
complexes, GnRH was docked into the catalytic channel of
both ACE domains, in the appropriate orientation for either
COOH- or NH2-terminal cleavage. Molecular dynamics
calculations have been employed as a means to allow for
enzyme reorganization in the presence of substrate. All
simulations were carried out with explicit solvent representation so as to examine the effect of solvent molecules in
mediating intermolecular interactions. This study extends our
knowledge of the structural determinants that contribute to
substrate specificity, which will aid the design of secondgeneration domain-specific ACE inhibitors.
COMPUTATIONAL METHODS
Structure Preparation. The crystal structure coordinates
of ACE-lisinopril complexes were obtained from RSC: pdb
code 1O86, for the C domain (tACE) and pdb code 2C6N,
for the N domain of sACE. All heteroatoms with the
exception of zinc and chloride ions were removed. Missing
atoms and hydrogen atoms were added using the LEaP
module of AMBER 8 (29). The protonation state of ionizable
groups was predicted by the program H++ using a continuum electrostatic model with the Poisson-Boltzmann
method (30). In addition to this, visual inspection of all
histidine residues for hydrogen bonding ability with their
neighboring residues was performed. The added atoms were
subjected to 500 rounds of energy minimization with steepest
descent gradients, while all other atoms were kept fixed by
applying 50 kcal‚mol-1‚Å-2 positional restraints. Implicit
solvent was applied using the generalized Born model GBHCT
(31) with 16 Å cutoff for the nonbonded interactions. GnRH
structure was also generated using LEaP and was subjected
to energy minimization in implicit solvent. For the pyroglutamic acid residue, GAFF parameters (32) and AM1-BCC
partial atomic charges (33) were assigned by the AMBER
module Antechamber. His2 residue of GnRH was assigned
its protonated form.
Docking of the Substrates. The program AutoDock 3.05
(34) was used for all docking calculations, and AutoDockTools was used for visual inspection of the docking results.
Protein and ligands were treated with the united-atom
approximation by merging all nonpolar hydrogens. Kollman
partial charges were assigned to all protein atoms, whereas
for zinc and chloride ions, formal charges and van der Waals
parameters from the AMBER database were assigned. The
Lamarckian genetic algorithm was employed with the
following standard parameters: a population size of 50
individuals; a maximum number of 1.5 × 106 energy
evaluations and a maximum number of 27,000 generations;
an elitism value of 1; a mutation rate of 0.02; and a crossover
rate of 0.80 (34). For most of the calculations, 100 docking
rounds were performed with step sizes of 0.25 Å for
translations and 5° for orientations and torsions. Docked
conformations were clustered with 1.5 Å tolerance for the
root-mean-square positional deviation.
Because of the relatively high conformational freedom of
the decapeptide substrate, its docking into the catalytic
channel of ACE with the scissile bond in the appropriate
orientation is practically difficult. For this reason, we have
performed the docking of GnRH1-5 and GnRH6-10 in two
steps, and their docked conformations were then linked
together. Two grid maps for each domain of ACE, one
centered at the S2-S1-S1′-S2′-S3′ subsites (zinc-binding site)
and the other at the S7-S6-S5-S4-S3 subsites (substrate
channel), were calculated using AutoGrid with 81 × 81 ×
81 grid points of 0.25 Å spacing. The two grid maps
overlapped at the S3-S2 subsite. In order to acquire the
GnRHC conformation, GnRH1-5 was docked into the substrate channel and GnRH6-10 at the zinc binding site and
vice versa for GnRHN. Selection of the conformations from
the docking calculations is based on different criteria for each
part of the substrate. When referring to the accepted
conformations at the ACE zinc-binding domain, two criteria
should be satisfied: (I) the distance between Leu7 (GnRHC)
or Trp3 (GnRHN) carbonyl oxygen and zinc is shorter than
3.0 Å, and (II) the N-terminus of Gly6 (GnRHC) or the
C-terminus of Tyr5 (GnRHN) is oriented toward the substrate
channel domain. For the substrate channel domain, a docked
conformation is accepted when the distance between the
Simulated Interactions of the ACE-GnRH Complex
C-terminus carbon of Tyr5 and the N-terminus nitrogen of
Gly6 is shorter than 4.0 Å.
However, the choice of an appropriate conformation for
each fragment is hampered by the extensive conformational
space of the pentapeptides. For instance, even after 100
rounds of docking Gly6-nGly10 into the zinc-binding domain
of nACE, there was no acceptable solution. The docked
conformations that fulfilled criterion I did not have the
expected orientation of the N-terminal group, whereas those
consistent with criterion II were at an inappropriate distance
from the zinc. Therefore, performing 200-300 rounds of
docking was necessary. In order to achieve an increased
number of conformations that fulfill criteria I and II, the
charge of Leu7O was changed from -0.50 to -1.50 e. In
this way, at least two accepted conformations of the Gly6nGly10 fragment at the zinc-binding domain of nACE were
obtained per 100 docking rounds. Docked conformations
were ranked by the predicted binding energy, though the final
choice was also based on visual inspection of the enzymesubstrate interactions, in order to verify that accommodation
of the P2-P1-P1′ side chains is reasonable according to the
X-ray crystal structures of both ACE domains with lisinopril
(see Results and Discussion).
After having selected the appropriate P2-P3′ conformation,
P2 and P1 moieties were included in the calculation of the
second grid, being regarded as part of the substrate channel.
Consequently, each predicted conformation of the P7-P3
fragment occupied distinct sites from those of the zincbinding domain fragment so that all accepted conformations
were productive. The top-ranked binding modes among the
accepted conformations were also visually inspected in order
to validate that intermolecular interactions were predicted
properly; in particular, whether H-bonding donor and acceptor atoms were correctly positioned and whether hydrophobic residues lay inside hydrophobic subsites of the
enzyme. In this fashion, the choice of the P7-P3 fragment
depends on both the value of the predicted binding energy
as well as its reasonable placement inside ACE, with the
latter influenced by the presence of the zinc-bound GnRH
moiety (see Results and Discussion). Subsequently, GnRH1-5
and GnRH6-10 were linked together, and a bond between
the oxygen of the scissile bond and zinc was created using
LEaP. At this point, a TIP3P water molecule was placed
between the scissile bond and Glu411/389 (cACE/nACE),
which was also linked to zinc. Each ACE-substrate complex
was subjected to energy minimization as described above
while constraining the movement of all enzyme atoms.
Torsional angle restraints were imposed when necessary to
keep the Tyr5-Gly6 peptidic bond of GnRH in the transconformation.
Molecular Dynamics. All calculations were carried out
using SANDER and PMEMD programs of AMBER 8 (29).
The force field of Cornell et al. (35) was employed with
full representation of solvent using the TIP3P water model
(36). Periodic boundary conditions were imposed by means
of the particle mesh Ewald method with an 8 Å limit for the
direct space sum. Numerical integration was performed with
a 2-fs time step, and all bonds involving hydrogen atoms
were constrained with SHAKE (37). Temperature and
pressure controls were imposed using a Berendsen-type
algorithm (38) with coupling constants of 1.0 and 5.0 ps,
respectively. Force field parameters for the zinc-binding sites
Biochemistry, Vol. 46, No. 30, 2007 8755
have been added as described below. The protein-substrate
complexes were immersed in isometric truncated octahedron
water boxes, constructed from a cubic box of ∼90 Å, and
an appropriate number of Na+ ions was added to neutralize
the system. The following procedure was used in order to
equilibrate the position of solvent molecules, the temperature,
and the pressure of the system: (a) energy minimization for
100 steps with steepest descent and 900 steps with conjugate
gradients was performed by imposing 50 kcal‚mol-1‚Å-2
positional restraints on the solute atoms; (b) temperature was
gradually increased at 300 K (in 50 K steps) within six
rounds of 5-ps constant volume dynamics, while solute atoms
were kept fixed by imposing 10 kcal‚mol-1‚Å-2 restraints;
(c) restraints were then released within 20 ps in the NVT at
300 K; and (d) the density of the simulation box was
increased from 0.90 to 1.05 g‚cm-3 within 150 ps of constant
pressure dynamics. Subsequently, production runs of 2,0003,000 ps were carried out at physiological conditions
(300 K, 1 atm) in the NPT ensemble. The translational centerof-mass motion was removed every 1000 steps, and trajectories were updated every 500 steps (viz. every ps). For visual
inspection and analysis of the MD trajectories, the program
VMD 1.8 (39) was used. Solvent-accessible and buried
surface areas were calculated by the program MSMS 2.5
(40) with a probe radius of 1.5 Å within VMD. The
Figures were also prepared with VMD and plots using the
program GRACE 5.1.
Zinc Force Field Parameters. The zinc coordination sphere
in native ACE comprises His383/361, His387/365, Glu411/
389 (C-/N-domain numbering), and a solvent molecule. In
accordance with a well-accepted proposal for thermolysin
(13, 14), ACE-substrate complexes involve an analogous
pentacoordinated metal center. As soon as the substrates are
positioned in the central channel of the enzyme, the carbonyl
oxygen of the scissile bond binds to zinc by displacing the
zinc-bound water toward Glu384/362. We initially experimented with a nonbonded metal representation, using the
parameters proposed by Stote and Karplus (41). However,
we found that during the equilibration period, water moved
toward the P2 side of the substrate. Consequently, Glu384/
362 came to within a distance of 2.0 Å from zinc and
remained there for the rest of the simulation (data not shown).
Such a tendency of zinc to be hexacoordinated during
unrestrained MD calculations has been also observed by
others using the nonbonded approach with the AMBER force
field (42, 43).
For this reason, we adopted the bonded approach with
explicit bonds between the metal and its ligands. Force field
parameters were derived from hybrid B3LYP density functional calculations using GAUSSIAN 98 (44). A model of
the active site (Figure 1 in Supporting Information), was
extracted from the high-resolution crystal structure of tACE.
The inhibitor was replaced by a water molecule opposite
Glu411, and N-methylacetamide was docked as the substrate.
The HEMGH...E zinc-binding motif was retained, except that
Met was truncated into Gly and Glu411 into butyrate.
Geometry optimizations were then carried out with loose
convergence criteria using the LANL2DZ basis, which was
shown to perform quite well for analogous systems (45). For
comparison, 3-21G*/ and 6-31G*/B3LYP geometry optimizations were also carried out. Analogous force fields by
others (46, 47) were taken into consideration in extracting
8756 Biochemistry, Vol. 46, No. 30, 2007
the bond and angle parameters shown in Table 1 of the
Supporting Information. All dihedral angle parameters
including zinc were set to zero. The partial atomic charge
of zinc was set to +0.75 e, and the standard AMBER charge
of the five ligand atoms was increased by +0.25 e (i.e., the
two histidine N2, the glutamate O1, the zinc-bound water
oxygen, and the carbonyl oxygen of the substrate). This
minimal charge distribution is in good agreement with partial
charges reported for zinc and its coordinating atoms (48, 49)
by using the RESP fit method (50). The same LennardJones parameters as for the docking calculations (R* )
1.10 Å, * ) 0.0125 kcal‚mol-1) were employed (46).
The additional set of AMBER force field parameters
(Table 1 in Supporting Information) was assessed during the
course of molecular dynamics simulations by examining the
geometry of the zinc coordination sphere of the ACE-GnRH
complex in both orientations. The plots of zinc-ligand bond
and angle values as a function of simulation time (Figures 2
and 3 in Supporting Information) exhibit a well-defined
geometry that is in agreement with that provided by the
B3LYP density functional calculations. A minor point is that
Zn-OW and Zn-N of histidine exhibit shorter bonds in
comparison with those extracted from DFT calculations by
∼0.2 and ∼0.1 Å, respectively. As far as the imidazole N8/
N9-Zn-O angles are concerned, depending on the system
under study they exhibit a fluctuation between 110° and 140°.
This observation is in accordance with our DFT calculations
when comparing results from different basis sets, which led
us to assign the same value for both angle parameters (i.e.,
θο ) 125°).
RESULTS AND DISCUSSION
For the sake of simplicity in making direct comparisons
between the two opposite orientations of GnRH in the S
primed and non-primed subsites of both ACE domains,
residues of GnRHN toward the N-terminus are numbered
P1′-P3′, and those toward the C-terminus are numbered P1P7, as shown in Table 1. For the same reason, tACE that
corresponds to the C domain of sACE is referred as cACE
and to the N domain of sACE as nACE.
Docking of the Substrate. Prediction of small molecule or
short peptide binding modes becomes quite straightforward
because of the available crystal structures of ACE with the
bound inhibitor lisinopril (15, 16). The binding subsites S1,
S1′, and S2′ are well characterized, and the differences at
the active sites between the two domains of ACE are
documented (16). However, obtaining models of ACE with
larger substrates, such as GnRH, introduces two major
difficulties: first, subsites S7-S2 and S3′ need to be identified,
and second, a decapeptide such as GnRH occupies a
relatively large conformational space inside the catalytic
groove of ACE. In conjunction with the enzyme’s plasticity
and reorganization upon binding of the substrate, prediction
of the interactions that dominate their complex is a challenging task. To overcome this, we have carried out a
fragment- and knowledge-based docking method followed
by molecular dynamics calculations. In the first stage of this
approach, a binding conformation of the substrate is obtained
by flexible docking of two peptidic fragments: one that
brackets the scissile bond, of which the carbonyl oxygen
binds zinc, and one for the remaining residues. In the case
Papakyriakou et al.
of GnRH, the two fragments comprise pGlu1-Tyr5 and
Gly6-nGly10 (Table 1). Each fragment is docked at two
overlapping binding sites, one centered at the zinc-binding
domain and the other centered at the central groove and
parallel to helices R1 and R2. In order to obtain the proper
GnRH orientation for carboxy terminal cleavage (GnRHC),
the Gly6-nGly10 fragment is docked at the zinc-binding site
and pGlu1-Tyr5 at the substrate channel. In the opposite
orientation for amino terminal cleavage (GnRHN), docking
is performed with Gly6-nGly10 at the substrate channel and
pGlu1-Tyr5 at the zinc-binding domain (Figure 1).
Even for the pentapeptide fragments of GnRH with 18
and 15 flexible torsions for pGlu1-Tyr5 and Gly6-nGly10,
respectively, the conformational space is still large enough
to be sampled adequately. Τhe possibility of applying
distance restraints between ACE and GnRH during the
docking procedure would have significantly increased the
number of accepted conformations. Given that distance
restraints between a ligand and a protein atom cannot be
imposed using AutoDock, the negative charge of the Leu7
carbonyl oxygen was increased, so as to achieve an increased
number of docked conformations with the appropriate
orientation of the scissile bond. It should be noted that the
increase of the negative charge of the ligand results in
overestimation of the calculated binding energies because
of the artificially increased electrostatic energy of the Zn...O
pair. This approach is only suitable to compensate for a weak
distance restraint between the two opposite charges. Indeed,
the number of accepted conformations of GnRHC at the zincbinding domain of cACE is increased significantly upon
changing the charge of Leu7O as well as the estimated
energies that are higher by ∼3 kcal‚mol-1 (Table 2). At this
point, the accepted conformations were evaluated by comparing the binding mode of P1-P2′ with that of the
crystallographic position of the inhibitor complexes. In this
way, use of the experimental data on the distinct S1-S2′
subsites of ACE render the choice of the structure to be used
for subsequent MD calculations simpler. Thus, the highest
affinity conformation that exhibits a reasonable binding mode
with respect to the cocrystalized inhibitor was chosen (Figure
2).
As far as the second GnRHC fragment is concerned,
docking of pGlu1-Tyr5 into the substrate channel of cACE
produced 12 accepted conformations per 100 docking rounds
(Table 2). However, nine of these exhibited major steric
hindrances with the Gly6-nGly10 fragment, which was also
the case for nACE for both GnRH orientations. For this
reason, after having selected the top-ranked conformation
of the zinc-bound fragment, residues P2-P1 were included
in the substrate channel grid box for the subsequent docking
calculations. Using this approach, all solutions for the P7P3 fragment are prevented from overlapping with the selected
P2-P1 moiety of the zinc-binding domain, as illustrated in
Figure 4 of the Supporting Information for the corresponding
GnRHC fragments. The calculated binding energies are higher
than those calculated in the absence of the P2-P1 fragment
by ∼1.0 kcal‚mol-1 (Table 2), a result that is attributed to
the increased electrostatic contribution of the P3‚‚‚P2 charged
termini. A similar approach has been applied by McCammon
and co-workers in the relaxed-complex scheme that permits
the design of potent drugs by combining two or three ligand
fragments with weak affinity (51). The higher affinity ligand
Simulated Interactions of the ACE-GnRH Complex
Biochemistry, Vol. 46, No. 30, 2007 8757
FIGURE 1: Cartoon of ACE C-domain (and cACE, upper panel) and N-domain (nACE, lower panel) with GnRH shown as yellow sticks,
zinc as gray, the zinc-bound water as red, and the buried chloride ions as green spheres. GnRH is shown either docked for carboxy terminal
cleavage (GnRHC, upper) or for amino terminal cleavage (GnRHN, lower).
Table 2: Comparison of the Results Obtained for GnRHC Docking to cACE between Two Approaches for Each of the Substrate Partsa
GnRH fragment
ACE domain
A.
-nGly
(Leu7O: -0.50 e)
A. Gly6-nGly10
(Leu7O: -1.50 e)
B. pGlu1-Tyr5
cACE zincbinding domain
cACE zincbinding domain
cACE channel
without Gly6-Leu7
cACE channel
with Gly6-Leu7
Gly6
10
B. pGlu1-Tyr5
no. of
accepted
conformations
mean
docked
energy
mean binding
energy
mean rmsd
7
-18.0 (3.7)
-11.8 (2.9)
5.2 (1.9)
12
-20.6 (3.6)
-15.0 (2.9)
5.0 (1.3)
12b
-15.9 (2.1)
-9.0 (1.9)
7.4 (1.4)
18
-16.9 (1.7)
-9.9 (1.5)
6.9 (1.2)
a
(A) Docking of the C-terminal zinc-binding fragment by using the Kollman charge or by increasing the negative charge of the Leu7 carbonyl
oxygen atom. (B) Docking of the N-terminal fragment including or excluding the Gly6-Leu7 moiety of the top-ranked C-terminal fragment. The
results are given per 100 docking rounds with the standard deviation in parenthesis; energies are in kcal‚mol-1 and rmsd from the top-ranked
solution in Å. b Nine of these are not meaningful because of steric hindrance at the zinc-bound docked conformations.
is used at the first phase of docking and is then regarded as
part of the enzyme, which was found to introduce specificity
in the orientation of the second ligand within a possible linker
distance to the first ligand, while excluding any unproductive
binding modes. In the case of ACE-GnRH complexes, the
conformation selected for the subsequent MD runs was
among the top-three ranked results and was within a
calculated binding energy range of 1.5 kcal‚mol-1. The
docking results obtained for each domain of ACE with GnRH
in both orientations are summarized in Table 2 of the
Supporting Information. Finally, the two GnRH fragments
were linked together, and after the addition of the zinc-bound
water molecule and the bonding of zinc with the scissile bond
carbonyl oxygen, the substrate was subjected to energy
minimization using the AMBER molecular mechanics force
field.
Molecular Dynamics Calculations. Because the initial
binding modes of GnRH were predicted by keeping the
enzyme atoms in their crystallographic positions and in the
absence of solvent molecules, we have employed molecular
dynamics (MD) calculations as a means to introduce the
effect of protein flexibility. Explicit solvent treatment was
used so as to observe interactions between ACE and GnRH
that are mediated by the polar solvent. In this way, it is
possible to monitor local motions of the enzyme in the
presence of the substrate at the nanosecond time scale and
investigate the structural characteristics that govern their
interaction. The scope of this study is to investigate the
8758 Biochemistry, Vol. 46, No. 30, 2007
FIGURE 2: Initial conformations of the substrate’s fragment P2-P1P1′-P2′ docked inside the two domains of ACE: (A) cACE-GnRHC,
(B) cACE-GnRHN, (C) nACE-GnRHC and (D) nACE-GnRHN.
For comparison, the inhibitor lisinopril is shown at its crystallographic position as yellow sticks, zinc is shown as a gray sphere,
and the zinc-bound water is in red.
enzyme-substrate interactions immediately prior to cleavage,
either at the COOH- or at the NH2-terminal tripeptides. For
this reason, a water molecule is coordinated to Zn(II) so that
it can attack the carbonyl carbon of the scissile bond. In order
to accurately represent the zinc binding environment, force
field parameters were extracted by carrying out DFT calculations on a model of the zinc-binding domain of ACE
including N-methylacetamide as the substrate. This new set
of AMBER force field parameters appears to be effective in
representing the pentacoordinated zinc-binding geometry
throughout MD calculations. The MD trajectories were
analyzed in order to identify important intermolecular
interactions by extracting their geometric features (distances
and angles) as a function of simulation time. The major
interactions between ACE and GnRH are summarized in
Table 3. Hydrogen-bonding interactions were monitored
using 3.5 Å as the distance cutoff and 120° as the angle
cutoff, whereas hydrophobic interactions were included for
a pair of carbon atoms at a distance shorter than 4.5 Å. Only
interactions that are present more than half of the simulation
time are considered, whereas water-mediated hydrogen bonds
should be observed at least at 25% of the trajectory frames.
OVerall Enzyme-Substrate Conformation. The four systems appear to be well equilibrated after the first 100 ps, as
evident from the plots of the root-mean-square deviation
(rmsd) of either CR atoms of ACE or all atoms of GnRH
from their initial coordinates and as a function of simulation
time (Figure 3). For each enzyme-substrate complex, the
average value and the standard deviation in Å were calculated
for the total simulation time; cACE: 1.48 (0.23) - GnRHC:
1.86 (0.18); cACE: 1.47 (0.25) - GnRHN: 2.23 (0.28);
nACE: 1.45 (0.19) - GnRHC: 2.80 (0.48); nACE: 1.64
(0.24) - GnRHN: 3.42 (0.45). Noticeably, the complex that
exhibits the highest deviation from the initial coordinates is
nACE with GnRHN, in contrast to both complexes of cACE
that display the lowest variations. However, this observation
does not indicate that cACE complexes with GnRH are more
stable, rather that the conformations of the substrate predicted
Papakyriakou et al.
by the docking calculations display less reorganization during
the MD phase.
The radius of gyration (Rγ) of ACE in each complex with
GnRH exhibits minute fluctuations during the course of the
simulations (Figure 4A), thus supporting the stability of the
overall compact structure of ACE. In remarkably good
agreement with the initial value of Rγ ) 23.62 Å for cACE
and Rγ ) 24.45 Å for nACE, the mean value calculated
throughout the total simulation time with the standard
deviation in parenthesis is 23.74 (0.06) for cACE-GnRHC,
23.86 (0.05) for cACE-GnRHN, 24.72 (0.08) for nACEGnRHC, and 24.69 (0.11) for nACE-GnRHN. This is further
supported by the plots of the accessible surface area of the
enzyme complexes, which exhibit fluctuations within 2% of
the initial value and are calculated to be 23404 (197) for
cACE-GnRHC, 23085 (254) for cACE-GnRHN, 25730
(340) for nACE-GnRHC, and 25847 (467) for nACEGnRHN in Å2 (Figure 4B). The buried surface area of GnRH
as a function of simulation time is shown in Figure 4C and
D and is calculated to be 1427 (25) for cACE-GnRHC, 1431
(32) for cACE-GnRHN, 1404 (37) for nACE-GnRHC, and
1332 (25) for nACE-GnRHN in Å2.
A hinge mechanism has been recently proposed for
substrate entry into the active site cleft of tACE (52), on the
basis of homology to human ACE2, which was crystallized
both in an open conformation without inhibitor and in an
inhibitor-bound closed conformation (pdb IDs: 1R4L and
1R42, respectively) (53). Normal-mode analysis revealed
intrinsic flexibility about the active site of tACE so that six
hinge regions could be identified: 98-125, 296-297, 400409, 434-439, 535-537, and 569-578 (tACE numbering).
In particular, region 98-125 lies close to the surface between
the R2 lid helix and R4; region 569-578 lies between the
conserved HEMGH motif and E411 that binds zinc so that
hinging about this region results in opening up the catalytic
site (52). The mobility of the simulated ACE-GnRH
complexes was investigated by calculating the B-factors of
the backbone CR atoms from their root-mean square fluctuations during the 2-ns MD simulation (Figure 4). A comparison between the two domains of the enzyme with the
orientations of the two substrates reveals that (i) region 98125/71-98 (cACE/nACE) exhibits quite high B-factors,
indicating that substrate binding into the catalytic cleft might
not influence the mobility of residues about which the hinge
opens; (ii) region 569-578 (cACE) exhibits low B-factors,
whereas the corresponding 547-556 region in nACE shows
some degree of flexibility that can be attributed to the loop
region between helix R25 and R26; (iii) in both domains of
ACE, the region between sheet β1 and the following loop
up to sheet β2 exhibits high flexibility (i.e., residues 150158 in cACE and 125-136 in nACE); (iv) in cACE, regions
292-299 (between helix R9 and R10) and 305-310 (before
helix R11) display more than double the B-factors for GnRHC
in comparison with those for GnRHN, whereas the corresponding region in nACE (269-278) exhibits comparable
high flexibility for both GnRH orientations; (v) in nACE,
region 530-538 exhibits higher flexibility for GnRHN than
for GnRHC, in contrast to the corresponding region 552560 in cACE that has equally low B-factors for both GnRH
orientations; and (vi) the region between helix R14 and sheet
β4 (320-327) in nACE appears to be more flexible in
GnRHC than in GnRHN.
Simulated Interactions of the ACE-GnRH Complex
Biochemistry, Vol. 46, No. 30, 2007 8759
Table 3: ACE Residues of Domains C and N that Exhibit Major Electrostatic Interactions with Both Orientations of GnRHa
ACE
S7
S6
S5
S4
S3
S2
S1
S1′
S2′
S3′
Cdomain
Lys118
Glu123
Glu403
Glu403
Asp358
Ala356
Ala354
Glu162
Lys511
Lys511
Arg522
Asp121
Met223
Phe570
Glu411
His410
Trp357
Tyr360
Phe391
WAT
Arg522
His353
Ser355
Val518
Asp377
His513
Ala354
Val380
Val380
Phe457
Phe519
Tyr523
His513
Trp59
Thr92
Val119
Leu122
Ndomain
GnRHC
Lys118
Ser222
Tyr213
Glu123
Tyr51
Tyr213
Glu123
Glu403
Trp59
Ala400
Glu403
Trp220
Ile204
Pro519
Arg522
Arg402
Tyr360
Arg522
Trp357
Phe391
Glu403
His410
Pro519
WAT
Trp357
His513
Phe512
Val518
Asp415
Asp453
His513
Val379
Val380
Phe457
Tyr523
Phe527
Glu162
His353
Glu376
Ala354
Val380
Phe512
GnRHN
Arg90
Gly382
Thr97
Tyr338
Asp336
Ala334
Ala332
His331
Gln259
-
GnRHC
Arg381
Ser548
Ala383
Trp201
Tyr338
Pro385
Arg500
Leu32
Thr97
Leu98
Tyr372
Gly382
Tyr372
Arg500
Trp335
Tyr338
Tyr369
Tyr369
Glu389
His331
Phe490
His491
Ser357
Thr358
Glu362
Asp393
His491
Lys489
Phe435
His491
Tyr498
Tyr501
Phe505
Ala94
Thr97
Arg381
Leu32
Tyr338
Glu389
CL2
Tyr369
His388
Ala334
Tyr338
Val36
Trp335
Ala334
CL2
Asp43
Lys364
Asn494
Arg500
Val329
Trp335
Phe490
His331
His491
WAT
Ala332
Asp393
Phe435
His491
Tyr501
Phe505
Glu431
Asp393
Phe505
Ala332
Gln355
Thr358
His331
GnRHN
a Residues that provide hydrophobic contacts are marked in bold, whereas those found to be water-mediated are underlined. The zinc-bound
water is designated WAT and the buried chloride ion CL2.
FIGURE 3: Root-mean-square deviation as a function of simulation time from the initial model of ACE CR atoms (upper plots) and GnRH
all atoms (lower plots) in the complex of cACE (left) or nACE (right) with GnRHC (black lines) or GnRHN (red lines).
Enzyme-Substrate Specific Interactions. By modeling of
a Phe-His-Leu tripeptide substrate using the tACElisinopril structure, Sturrock et al. (54) have demonstrated
that (i) Tyr523 promotes the formation of the intermediate
similar to the role of His231 and Tyr157 in thermolysin by
forming a hydrogen bond between the hydroxyl group and
the scissile bond carbonyl oxygen; (ii) Ala354 plays a role
similar to that of Ala113 in thermolysin by stabilizing the
scissile bond nitrogen of the substrate via hydrogen-bonding
interaction with its carbonyl oxygen. An examination of the
calculated distances between the atoms of the GnRH scissile
bond and either Tyr523/501 or Ala354/332 of cACE/nACE
reveals that (i) the Tyr523/501 hydroxyl group is restricted
at a distance of 3.5-4.5 Å from the scissile bond oxygen of
both GnRH orientations and both ACE domains (Figure 5A
in Supporting Information) and that (ii) although Ala354/
332 is well positioned to interact with Arg8NH in GnRHC,
this is not the case for Ser4NH in GnRHN, which is removed
8760 Biochemistry, Vol. 46, No. 30, 2007
Papakyriakou et al.
FIGURE 4: Time dependence of the radius of gyration, Rγ (A), accessible surface area of the enzyme, ASA (B), and buried surface area
of the substrate, BSA (C, D) during the MD simulations of the ACE complexes with either GnRHC (black lines) or GnRHN (red lines).
from the carbonyl oxygen of either Ala354 in cACE or
Ala332 in nACE (Figure 5B in Supporting Information). This
observation implicates the placement of GnRHN scissile bond
for primary amino-terminal endoproteolytic cleavage by ACE
and raises the point whether a water-mediated interaction
between Ala354/332 and Ser4NH can compensate for a direct
hydrogen bond.
Another interesting observation is that the carbonyl oxygen
of either Ser4 in GnRHN or Gly6 in GnRHC forms a stable
hydrogen bond with the zinc-bound water molecule of cACE.
In contrast, the MD simulation of nACE reveals that only
Ser4O (GnRHN) and not Gly6O (GnRHc) is within proper
hydrogen-bonding distance (Figure 6). Gly6 of GnRHC
exhibits a hydrogen bond with the Tyr369 hydroxyl group
of nACE, a residue that is replaced by Phe391 in cACE.
The latter exhibits van der Waals contacts with the neighboring Tyr5 residue of GnRHC (Table 3). This observation
implies that the carbonyl oxygen of the P2-P1 peptide bond
(Ser4O in GnRHN or Gly6O in GnRHC) might contribute to
the enhancement of the zinc-bound water nucleophilicity via
a hydrogen-bonding interaction, in conjunction with the
polarization between zinc and Glu384/362 carboxylate that
is proposed to promote water’s attack on the carbon of the
scissile bond (14). The absence of such an interaction in the
nACE-GnRHC complex may be a factor that predisposes
the endoproteolytic activity of nACE to NH2-terminal
substrate cleavage.
Interactions that Support GnRHC CleaVage by Both ACE
Domains. By carrying out a series of sophisticated kinetic
measurements on a variety of ACE substrates, Husain and
co-workers have recently shed light on the basis of the
exopeptidase activity of the C domain (55). This mainly
involves the carboxylate-docking interactions with Lys511
and Tyr520, as was previously observed in the tACElisinopril crystal structure (15). Because GnRH is protected
at both NH2- and COOH-termini, such carboxylate-docking
interactions become less important for transition-state stabilization. Analysis of our MD data reveals that the conserved
Lys511/489 residue participates in the binding of GnRHC
by forming a hydrogen bond with the Pro9 backbone, in
contrast to GnRHN, where S2′ is occupied by the protonated
His2 and exhibits no interactions with Lys511/489. Concerning Tyr520/498, it appears to be involved only in the
interaction of GnRHC with the S2′ subsite of nACE (Table
3). Therefore, cleavage of the C-terminal tripeptide of GnRH
FIGURE 5: B-factors calculated from the atomic fluctuations of ACE
CR atoms during the course of the 2-ns MD trajectory (red) in
comparison with the corresponding crystallographic values (black).
(A) cACE-GnRHC, (B) cACE-GnRHN, (C) nACE-GnRHC, and
(D) nACE-GnRHN complexes.
is not expected to be strongly dependent on these two ACE
residues that provide the carboxylate-docking interactions.
However, major electrostatic interactions are displayed at
the S1′ that is occupied by Arg8 of GnRHC. In cACE, both
Glu162 and Asp377 are predicted to form salt bridges with
Arg8, similar to Glu362 and Asp393 in nACE. Additional
hydrogen-bonding interactions are provided by the conserved
His513/491 residue of both ACE domains, in addition to
Ser357 and Thr358 of nACE (Table 3). These interactions
are thus expected to contribute significantly to the endoproteolytic cleavage of the GnRH C-terminus by both domains
Simulated Interactions of the ACE-GnRH Complex
FIGURE 6: Representation of the chloride binding site of the nACE
(cyan sticks)-GnRHN (yellow sticks) complex. The Cl2 ion is
shown as a green sphere, the zinc ion is in gray, and the zincbound water is in red.
of ACE. Furthermore, the S2′ subsite in both ACE domains
accommodates Pro9 of GnRHC into a hydrophobic cage of
aromatic rings comprising Phe and Tyr residues (Figure 7B
and D). This is also the case for Leu7, which is well
positioned to exhibit both hydrophobic and hydrogenbonding interactions at the S1 subsite of ACE.
Apart from the interactions observed near the zinc-binding
site, the S3-S7 subsites provide major interactions with
GnRHC. The P5 residue Trp3 of GnRHC interacts with the
chloride-binding Arg522/500 residue of cACE/nACE and
exhibits hydrophobic interactions with Phe570 and Met223
of cACE or Leu32 and Leu98 of nACE. Interestingly, the
S3 subsite provides major interaction with Tyr5 of GnRHC
in contrast to GnRHN where Gly6 exhibits fewer interactions
with either domain of ACE. The conserved Asp358/336
residue backbone forms stable hydrogen bonds with Tyr5,
whereas Tyr372 of nACE provides further stabilization with
GnRHC (Table 3). The conserved Trp357/335 of ACE and
Phe391 of cACE or the corresponding Tyr369 of nACE
exhibit van der Waals contacts with the Tyr5 side chain.
Finally, the S7 subsite with Lys118 in cACE or Arg90 and
Arg381 in nACE provide the appropriate context for interaction with the polar groups of pGlu1 in GnRHC. In addition,
a hydrophobic cage comprising Trp59, Val119, and Leu122
accommodates pGlu1 in cACE, interactions that were not
detected in the simulations of nACE-GnRHC. However,
pGlu1 along with Trp3 and Tyr5 of GnRH orientation for
COOH-terminal cleavage displays similar interactions with
both domains of ACE.
Interactions that Support GnRHN CleaVage by nACE.
Remarkably, in all MD trajectories, the conserved Arg522/
500 residue of cACE/nACE exhibits either direct or watermediated hydrogen-bonding interactions with GnRH
(Table 3). This residue is one of the three Cl2 ligands
(Tyr224/202 and a water molecule are the other two) and is
reported to be critical for the chloride dependence of ACE
activity (56). Our results indicate that there is a direct
interaction between both Arg522 and Arg500 and Tyr5 of
Biochemistry, Vol. 46, No. 30, 2007 8761
GnRHN, whereas for GnRHC, Trp3 exhibits water-mediated
hydrogen bonds with Arg522 in cACE or Arg500 in nACE.
This finding is in agreement with a previous suggestion that
this residue, which lies on the same helix (R17 of tACE) as
that of two residues (Tyr520 and Tyr523) that interact with
the inhibitor, may interact with the substrate as well (15).
What is of more interest is that only in the simulation of
the nACE-GnRHN complex does the guanidine group of
Arg8 display a strong electrostatic interaction with the Cl2
ion, a contact that is not present in the initial docked
conformation.ThisinteractionresultsinannACEArg500+‚‚‚Cl-‚‚‚+Arg8
substrate bridge, which can increase the stability of the
ground and transition states (Figure 6). In addition to this,
Arg8 is found to stabilize the zinc-binding Glu389 through
a hydrogen bond between the guanidine and carboxylate
(Figure 6 in Supporting Information), an interaction that
might further increase the affinity of the enzyme-substrate
complex. At this point, it must be noted that Tyr202 (the
other Cl2 ligand in nACE) is displaced upon the interaction
of Arg8 with the chloride ion.
On the basis of an earlier report that the affinity of chloride
binding to ACE is increased with substrates that contain a
basic P2′ side chain, particularly that of arginine (19), an
analogous bridging interaction has been recently proposed
for cACE and P2′ Arg (55). From our data, the protonated
His2 (P2′) residue of GnRHN does not appear to be implicated
in such an interaction; however, this hypothesis seems to be
plausible on a structural basis for a basic residue in a position
other than the P2′, for example, Arg8 of GnRHN. Taking into
account that in all the other MD simulations of GnRH with
ACE (Table 3) there is no direct interaction with Cl2, this
ionic bridge might play a major role in the endoproteolytic
cleavage of GnRH by nACE, which is in agreement with
experimental evidence that the N domain of ACE is mainly
responsible for the primary NH2-terminal cleavage of GnRH
(24, 28).
Recently, Sturrock and co-workers synthesized a lisinopril-tryptophan analogue inhibitor of ACE (57) on the basis
of the interactions of the phosphonic acid inhibitor RXPA380
(23). Molecular modeling studies have revealed that the
indole-NH has the ability to establish a strong hydrogen
bond with Asp453 of cACE and that the tryptophan makes
hydrophobic interactions with Val379 and Val380. This is
also apparent for Trp3 of GnRHN, which displays hydrogenbonding interactions with Asp393 at the S1′ of nACE (Table
3), in contrast to cACE for which such an interaction is not
observed (Figure 7B and D). Therefore, the P1′ residue of
GnRHN is predicted to contribute appreciably to the specificity of nACE for NH2-terminal cleavage.
Interactions that Contribute to the Affinity of the ACEGnRH Complex. A number of important interactions between
ACE and GnRH cannot be strictly classified as described
above but are predicted to contribute significantly to the in
Vitro endoproteolytic activity of ACE. The P2′ basic histidine
residue of GnRHN exhibits strong electrostatic interactions
with both domains of ACE, if it is protonated. Specifically,
Asp415 and Asp453 residues of cACE or Asp393 and
Glu431 of nACE form hydrogen bonds with the His2
imidazole ring. The conserved Phe527/505 residue exhibits
van der Waals contacts with His2 of GnRHN in both ACE
domains, whereas Val379 and Val380 of cACE that exhibit
contacts with GnRHN are replaced by the polar Ser357 and
8762 Biochemistry, Vol. 46, No. 30, 2007
Papakyriakou et al.
FIGURE 7: S prime subsites of (A) cACE-GnRHC, (B) cACE-GnRHN, (C) nACE-GnRHC, and (D) nACE-GnRHN complexes;
representations are as in Figure 6.
Thr358 residues in the S1′ subsite of nACE. As far as the
S3′ subsite is concerned, pGlu1 of GnRHN exhibits electrostatic and hydrophobic interactions with both domains of
ACE, though cACE is predicted to provide more interactions
directly (Table 3). On the basis of these observations, it is
predicted that P2′ and P3′ residues might contribute equally
to the affinity of both ACE domains for GnRHN.
Even though Gly6 exhibits a few contacts in both GnRH
orientations, this residue is predicted to mediate some key
interactions. Particularly for GnRHC, Gly6 provides additional
stabilization to the zinc-bound water of cACE, whereas for
nACE, it is implicated in two hydrogen bonds with Ala334
and Tyr369 (Table 3). At the opposite orientation, Gly6 of
GnRHN is predicted to interact with the chloride-binding site,
either directly with Arg522 of cACE or by a water-mediated
interaction with the Cl2 ion of nACE. As far as the S4 subsite
is concerned, Glu403 of cACE exhibits interactions with
either the Ser4 hydroxyl group of GnRHC or the Leu7
backbone of GnRHN, whereas in nACE, this residue is
replace by Tyr338. In addition, Leu7 of GnRHN makes
hydrophobic contacts with Trp220 and Ile204 of cACE or
Val36 and Trp335 of nACE (Table 3).
As discussed above, Arg8, which occupies the S5 subsite
of the nACE-GnRHN complex, is able to bind the Cl2 ion.
In cACE, Arg8 is predicted to be implicated in a salt bridge
with Glu123, a residue that can also provide electrostatic
interactions with His2 of GnRHC or the Pro9 carboxylate
group of GnRHN at the S6 subsite of cACE (Table 3). In
nACE, this residue is replaced by Gly99, which cannot
mediate similar interactions; however, Gly382 and Ala383
display hydrogen bonds with His2 of GnRHC and Arg381
with Pro9 of GnRHN. Further stabilization through hydrophobic contacts with His2 is provided by Trp201 and Tyr338
of nACE, in contrast to cACE for which no such interactions
were observed. Pro9 makes van der Waals contacts with
Tyr213 of cACE or Leu32 and Tyr338 of nACE (Table 3).
As far as the protected termini of GnRH are concerned,
nGly10 of GnRHN displays hydrogen-bonding interactions
with Lys118, Ser222, and Tyr213 at the S7 of cACE, or
Ala94 and Thr97 of nACE. In the opposite orientation, nGly10
Simulated Interactions of the ACE-GnRH Complex
on GnRHC reveals minor interactions with both ACE
domains at the S3′ subsites, which are mainly water-mediated
hydrogen bonds. In contrast, when the S3′ subsite is occupied
by pGlu1 of GnRHN, major interactions are exhibited with
both ACE domains and especially with cACE (Table 3). This
residue was also predicted to exhibit strong electrostatic and
hydrophobic interactions at the S7 subsite of the cACEGnRHC complex.
CONCLUSIONS
Although the precise biological role of ACE in the
processing of GnRH is unclear, there are several studies that
suggest that it might be important for the degradation of the
hormone. Alternative peptidases are capable of hydrolyzing
the C-terminal residues from prohormone intermediates in
mice lacking carboxypeptidase activity (58). Among these,
somatic ACE efficiently removes the C-terminal dipeptide
from the Gly-Lys-Arg-extended GnRH. Furthermore, enzymatic degradation of GnRH by intestinal mucosal tissues is
regulated by endopeptidase-24.18 (EC 3.4.24.18), endopeptidase-24.15 (EC 3.4.24.15), and ACE as indicated by the
formation of GnRH(1-3) and GnRH(1-4) (59). Finally, the
activity of GnRH-inactivating peptidases may vary in different reproductive states such as across the estrous cycle,
and recently, Shimizu et al. showed the altered expression
of ACE mRNA during the periovulatory phase in GnRHtreated cows (60). Thus, the elucidation of the mechanism
of how ACE cleaves this important hormone is key to our
understanding of ACE’s possible role in the endocrine
system.
A series of docking calculations in combination with MD
simulations have been carried out in order to study the
structural characteristics of an ACE-GnRH complex. The
substrate was docked inside the catalytic groove of both
domains of ACE in appropriate conformation for either
COOH- or NH2-terminal endoproteolytic cleavage, and the
reorganization of their complex was monitored by MD
calculations. Although both domains of ACE provide the
appropriate context for either direct or water-mediated
interactions with the substrate, our data indicate that the N
domain of ACE provides an additional electrostatic interaction with the substrate, which may explain its preference for
the primary amino-terminal endoproteolytic cleavage of
GnRH in Vitro. On a structural basis, it is possible that the
Arg8 guanidine side chain of GnRHN interacts directly with
the chloride ion of nACE as well as stabilizes the zinc-bound
Glu389 carboxylate through hydrogen-bonding interaction.
Such an interface is predicted to comprise an nACE
Arg500+‚‚‚Cl-‚‚‚+Arg8 GnRHN bridging interaction. In support of this observation, the protected N-terminal pGlu1
residue of GnRHN exhibits stronger interactions in the S3′
subsite of both ACE domains, with respect to those provided
by the amidated C-terminal nGly10 of GnRHC. By virtue of
the absence of any C-terminal carboxylate-docking interactions with the protected peptide, the contribution of pGlu1
in stabilizing the ground and transition states of the enzymesubstrate complex may play a key role.
Moreover, apart from the carboxylate group of Glu384/
362 in cACE/nACE that polarizes the zinc-bound water that
attacks the carbon of the scissile bond, Ser4 or the Gly6
carbonyl oxygen of GnRHN or GnRHC, respectively, exhibits
Biochemistry, Vol. 46, No. 30, 2007 8763
additional stabilization through a hydrogen bond with the
zinc-bound water. This has been observed in all MD
simulations with the exception of that of the nACE-GnRHC
complex. However, both orientations of GnRH display
hydrogen-bonding interactions with the conserved Arg522/
500 chloride-binding residue, either directly or in a watermediated manner, interactions that support the basis of the
chloride-dependent activity of ACE.
In conclusion, this simulation study of a zinc-metalloprotease complex with a peptide substrate reveals valuable
information on the structural determinants of their interaction
and offers new insights into the dynamic nature of their
binding. The data reported here provide a structural basis
for the chloride-dependent activity of ACE and an appreciation of new enzyme-substrate contacts that contribute to
substrate recognition. In addition, these simulations provide
an extended view of the catalytic subsites that are more
distant from the metal center in comparison with those
identified by the small-molecule inhibitors. These findings
are likely to contribute to the targeting of the distal binding
subsites in the design of next-generation domain-selective
inhibitors of ACE with improved pharmacological profiles.
ACKNOWLEDGMENT
We thank the referees for their constructive comments on
the manuscript. A.P. is grateful to Yannis Lazarou for his
assistance with the Gaussian calculations and to David Case
for providing the AMBER software.
NOTE ADDED AFTER ASAP PUBLICATION
This article was released ASAP on July 3, 2007, with a
minor error in one of the pdb codes in the first paragraph of
the Computational Methods section. The correct version was
posted on July 13, 2007.
SUPPORTING INFORMATION AVAILABLE
A figure with the model of the catalytic site of ACE used
for the DFT calculations, a table with important geometrical
features of the zinc coordination sphere and the derived
AMBER force field parameters, four figures with plots of
selected distances and angles versus simulation time of ACE
with GnRH, a table summarizing the docking results of each
GnRH fragment with both ACE domains, and a figure
illustrating the highest docked conformations of the cACEGnRHC complex. This material is available free of charge
via the Internet at http://pubs.acs.org.
REFERENCES
1. Kearney, P. M., Whelton, M., Reynolds, K., Muntner, P., Whelton,
P. K., and He, J. (2005) Global burden of hypertension: analysis
of worldwide data, Lancet 365, 217-223.
2. Inagami, T. (1994) The renin-angiotensin system, Essays Biochem.
28, 147-164.
3. Turner, A. J., and Hooper, N. M. (2002) The angiotensinconverting enzyme gene family: genomics and pharmacology,
Trends Pharmacol. Sci. 23, 177-183.
4. Patchett, A. A., Harris, E., Tristram, E. W., Wyvratt, M. J., Wu,
M. T., Taub, D., Peterson, E. R., Ikeler, T. J., ten Broeke, J., Payne,
L. G., Ondeyka, D. L., Thorsett, E. D., Greenlee, W. J., Lohr, N.
S., Hoffsommer, R. D., Joshua, H., Ruyle, W. V., Rothrock, J.
W., Aster, S. D., Maycock, A. L., Robinson, F. M., Hirschmann,
R., Sweet, C. S., Ulm, E. H., Gross, D. M., Vassil, T. C., and
Stone, C. A. (1980) A new class of angiotensin-converting enzyme
inhibitors, Nature 288, 280-283.
8764 Biochemistry, Vol. 46, No. 30, 2007
5. Zaman, M. A., Oparil, S., and Calhoun, D. A. (2002) Drugs
targeting the renin-angiotensin-aldosterone system, Nat. ReV. Drug
DiscoVery 1, 621-636.
6. Acharya, K. R., Sturrock, E. D., Riordan, J. F., and Ehlers, M. R.
(2003) ACE revisited: a new target for structure-based drug
design, Nat. ReV. Drug DiscoVery 2, 891-969.
7. Turk, B. (2006) Targeting proteases: successes, failures and future
prospects, Nat. ReV. Drug DiscoVery 5, 785-799.
8. Barrett, A. J. (1998) Introduction: other Families in Clan MA, in
Handbook of Proteolytic Enzymes (Barret, A. J., Rawlings, N.
D., and Woessner, J. F., Eds.), pp 1033-1037, Academic Press,
New York.
9. Corvol, P., Williams, T. A., and Soubrier, F. (1995) Peptidyl
dipeptidase A: angiotensin I-converting enzyme, Methods Enzymol. 248, 283-305.
10. Hubert, C., Houot, A. M., Corvol, P., and Soubrier, F. (1991)
Structure of the angiotensin I-converting enzyme gene. Two
alternate promoters correspond to evolutionary steps of a duplicated gene, J. Biol. Chem. 266, 15377-15383.
11. Ehlers, M. R., Fox, E. A., Strydom, D. J., and Riordan, J. F. (1989)
Molecular cloning of human testicular angiotensin-converting
enzyme; the testis isoenzyme is identical to the C-terminal half
of endothelial angiotensin-converting enzyme, Proc. Natl. Acad.
Sci. U.S.A. 86, 7741-7745.
12. Soubrier, F., Alhenc-Gelas, F., Hubert, C., Allegrini, J., John, M.,
Tregear, G., and Corvol, P. (1988) Two putative active centers in
human angiotensin I-converting enzyme revealed by molecular
cloning, Proc. Natl. Acad. Sci. U.S.A. 85, 9386-9390.
13. Matthews, B. W. (1988) Structural basis of the action of
thermolysin and related zinc peptidases, Acc. Chem. Res. 21, 333340.
14. Hangauer, D. G., Monzinqo, A. F., and Matthews, B. W. (1984)
An interactive computer graphics study of thermolysin-catalyzed
peptide cleavage and inhibition by N-carboxymethyl dipeptides,
Biochemistry 23, 5730-5741.
15. Natesh, R., Schwager, S. L. U., Sturrock, E. D., and Acharya, K.
R. (2003) Crystal structure of the human angiotensin-converting
enzyme-lisinopril complex, Nature 421, 551-554.
16. Corradi, H. R., Schwager, S. L. U., Nchinda, A. T., Sturrock, E.
D., and Acharya, K. R. (2006) Crystal structure of the N domain
of human somatic angiotensin I-converting enzyme provides a
structural basis for domain-specific inhibitor design, J. Mol. Biol.
357, 964-974.
17. Redelinghuys, P., Nchinda, A. T., and Sturrock, E. D. (2005)
Development of domain-selective angiotensin I-converting enzyme
inhibitors, Ann. N.Y. Acad. Sci. 1056, 160-175.
18. Spyroulias, G. A., and Cordopatis, P. (2005) Current inhibition
concepts of zinc metallopeptidases involved in blood pressure
regulation, Curr. Enz. Inhibition 1, 29-42.
19. Shapiro, R., Holmquist, B., and Riordan, J. F. (1983) Anion
activation of angiotensin converting enzyme: dependence on
nature of substrate, Biochemistry 22, 3850-3857.
20. Natesh, R., Schwager, S. L. U., Evans, H. R., Sturrock, E. D.,
and Acharya, K. R. (2004) Structural details on the binding of
antihypertensive drugs captopril and enalaprilat to human testicular
angiotensin I-converting enzyme, Biochemistry 43, 8718-8724.
21. Wei, L., Clausen, E., Alhenc-Gelas, F., and Corvol, P. (1992) The
two homologous domains of angiotensin I-converting enzyme
interact differently with competitive inhibitors, J. Biol. Chem. 267,
13398-13405.
22. Dive, V., Cotton, J., Yiotakis, A., Michaud, A., Vassiliou, S.,
Jiracek, J., Vazeux, G., Chauvet, M. T., Cuniasse, P., and Corvol,
P. (1999) RXP407, a phosphonic peptide, is a potent inhibitor of
angiotensin I converting enzyme able to differentiate between its
two active sites, Proc. Natl. Acad. Sci. U.S.A. 96, 4330-4335.
23. Georgiadis, D., Cuniasse, P., Cotton, J., Yiotakis, A., and Dive,
V. (2004) Structural determinants of RXPA380, a potent and
highly selective inhibitor of the angiotensin-converting enzyme
C-domain, Biochemistry 43, 8048-8054.
24. Jaspard, E., Wei, L., and Alhenc-Gelas, F. (1993) Differences in
the properties and enzymatic specificities of the two active sites
of angiotensin-I converting enzyme (kininase II). Studies with
bradykinin and other natural peptides, J. Biol. Chem. 268, 94969503.
25. Deddish, P. A., Marcic, B., Jackman, H. L., Wang, H. Z., Skidgel,
R. A., and Erdos, E. G. (1998) N-domain-specific substrate and
C-domain inhibitors of angiotensin-converting enzyme: angiotensin-(1-7) and keto-ACE, Hypertension 31, 912-917.
Papakyriakou et al.
26. Rousseau, A., Michaud, A., Chauvet, M. T., Lenfant, M., and
Corvol, P. (1995) The hemoregulatory peptide N-Acetyl-Ser-AspLys-Pro is a natural and specific substrate of the N-terminal active
site of human angiotensin-converting enzyme, J. Biol. Chem. 270,
3656-3661.
27. Skidgel, R. A., and Erdos, E. G. (1985) Novel activity of human
angiotensin I converting enzyme: release of the NH2- and COOHterminal tripeptides from the luteinizing hormone-releasing hormone. Proc. Natl. Acad. Sci. U.S.A. 82, 1025-1029.
28. Ehlers, M. R., and Riordan, J. F. (1991) Angiotensin-converting
enzyme: zinc- and inhibitor-binding stoichiometries of the somatic
and testis isoenzymes, Biochemistry 30, 7118-7126.
29. Case, D. A., Darden, T. A., Cheatham III, T. E., Simmerling, C.
L., Wang, J., Duke, R. E., Luo, R., Merz, K. M., Jr., Wang, B.,
Pearlman, D. A., Crowley, M., Brozell, S., Tsui, V., Gohlke, H.,
Mongan, J., Hornak, V., Cui, G., Beroza, P., Schafmeister, C.,
Caldwell, J. W., Ross, W. S., and Kollman P. A. (2004) AMBER
8, University of California, San Francisco, CA.
30. Bashford, D., and Karplus, M. (1990) pKa of ionizable groups in
proteins: atomic detail from a continuum electrostatic model,
Biochemistry 29, 10219-10225.
31. Tsui, V., and Case, D. A. (2001) Theory and applications of the
generalized Born solvation model in macromolecular simulations,
Biopolymers 56, 275-291.
32. Wang, J., Wolf, R. M., Caldwell, J. W., Kollman, P. A., and Case,
D. A. (2004) Development and testing of a general Amber force
field, J. Comput. Chem. 25, 1157-1174.
33. Jakalian, A., Jack, D. B., and Bayly, C. I. (2002) Fast, efficient
generation of high-quality atomic charges. AM1-BCC model: II.
Parameterization and validation, J. Comput. Chem. 23, 16231641.
34. Morris, G. M., Goodsell, D. S., Halliday, R. S., Huey, R., Hart,
W. E., Belew, R. K., and Olson, A. J. (1998) Automated docking
using a Lamarckian genetic algorithm and an empirical binding
free energy function, J. Comput. Chem. 19, 1639-1662.
35. Cornell, W. D., Cieplak, P., Bayly, C. I., Gould, I. R., Merz, K.
M., Jr., Ferguson, D. M., Spellmeyer, D. C., Fox, T., Caldwell, J.
W., and Kollman, P. A. (1995) A second generation force field
for the simulation of proteins, nucleic acids, and organic molecules, J. Am. Chem. Soc. 117, 5179-5197.
36. Jorgensen, W. L., Chandrasekhar, J., Madura, J., and Klein, M.
L. (1983) Comparison of simple potential functions for simulating
liquid water, J. Chem. Phys. 79, 926-935.
37. Miyamoto, S., and Kollman, P. A. (1992) Settle: an analytical
version of the SHAKE and RATTLE algorithm for rigid water
models, J. Comput. Chem. 13, 952-962.
38. Berendsen, H. J. C., Postma, J. P. M., van Gunsteren, W. F.,
DiNola, A., and Haak, J. R. (1984) Molecular dynamics with
coupling to an external bath, J. Chem. Phys. 81, 3684-3690.
39. Humphrey, W., Dalke, A., and Schulten, K. (1996) VMD: visual
molecular dynamics, J. Mol. Graphics 14, 33-38.
40. Sanner, M. F., Spehner, J.-C., and Olson, A. J. (1996) Reduced
surface: an efficient way to compute molecular surfaces, Biopolymers 38, 305-320.
41. Stote, R. H., and Karplus, M. (1995) Zinc binding in proteins and
solution: a simple but accurate nonbonded representation, Proteins
23, 12-31.
42. Donini, O. A. T., and Kollman, P. A. (2000) Calculation and
prediction of binding free energies for the matrix metalloproteinases, J. Med. Chem. 43, 4180-4188.
43. Merz, K. M., Jr. (1991) C02 Binding to human carbonic anhydrase
II, J. Am. Chem. Soc. 113, 406-411.
44. Frisch, M. J., Trucks, G. W., Schlegel, H. B., Scuseria, G. E.,
Robb, M. A., Cheeseman, J. R., Zakrzewski, V. G., Montgomery,
J. A., Jr., Stratmann, R. E., Burant, J. C., Dapprich, S., Millam,
J. M., Daniels, A. D., Kudin, K. N., Strain, M. C., Farkas, O.,
Tomasi, J., Barone, V., Cossi, M., Cammi, R., Mennucci, B.,
Pomelli, C., Adamo, C., Clifford, S., Ochterski, J., Petersson, G.
A., Ayala, P. Y., Cui, Q., Morokuma, K., Malick, D. K., Rabuck,
A. D., Raghavachari, K., Foresman, J. B., Cioslowski, J., Ortiz,
J. V., Stefanov, B. B., Liu, G., Liashenko, A., Piskorz, P.,
Komaromi, I., Gomperts, R., Martin, R. L., Fox, D. J., Keith, T.,
Al-Laham, M. A., Peng, C. Y., Nanayakkara, A., Gonzalez, C.,
Challacombe, M., Gill, P. M. W., Johnson, B. G., Chen, W.,
Wong, M. W., Andres, J. L., Head-Gordon, M., Replogle, E. S.,
Pople, J. A. (1998) Gaussian 98, revision A.9, Gaussian, Inc.,
Pittsburgh, PA.
Simulated Interactions of the ACE-GnRH Complex
45. Pelmenschikov, V., Blomberg, M. R. A., and Siegbahn, P. E. M.
(2002) A theoretical study of the mechanism for peptide hydrolysis
by thermolysin, J. Biol. Inorg. Chem. 7, 284-298.
46. Hoops, S. C., Anderson, K. W., and Merz, K. M., Jr. (1991) Force
field design for metalloproteins, J. Am. Chem. Soc. 113, 82628270.
47. Tuccinardi, T., Martinelli, A., Nuti, E., Carelli, P., Balzano, F.,
Uccello-Barretta, G., Murphy G., and Rossello, A. (2006) Amber
force field implementation, molecular modelling study, synthesis
and MMP-1/MMP-2 inhibition profile of (R)- and (S)-N-hydroxy2-(N-isopropoxybiphenyl-4-ylsulfonamido)-3-methylbutanamides, Bioorg. Med. Chem. 14, 4260-4276.
48. Park, H., and Merz, K. M., Jr. (2005) Force field design and
molecular dynamics simulations of the carbapenem- and cephamycin-resistant dinuclear zinc metallo-β-lactamase from Bacteroides fragilis and its complex with a biphenyl tetrazole inhibitor, J.
Med. Chem. 48, 1630-1637.
49. Park, H., and Lee, S. (2004) Homology modeling, force field
design, and free energy simulation studies to optimize the activities
of histone deacetylase inhibitors, J. Comput.-Aided Mol. Des. 18,
375-388.
50. Bayly, C. I., Cieplak, P., Cornell, W. D., and Kollman, P. A. (1993)
A well-behaved electrostatic potential based method using charge
restraints for deriving atomic charges: the RESP model, J. Phys.
Chem. 97, 10269-10280.
51. Lin, J.-H., Perryman, A. L., Schames, J. R., and McCammon, A.
J. (2002) Computational drug design accommodating receptor
flexibility: the relaxed complex scheme, J. Am. Chem. Soc. 124,
5632-5633.
52. Watermeyer, J. M., Sewell, B. T., Schwager, S. L., Natesh, R.,
Corradi, H. R., Acharya, K. R., and Sturrock, E. D. (2006)
Structure of testis ACE glycosylation mutants and evidence for
conserved domain movement, Biochemistry 45, 12654-12663.
53. Towler, P., Staker, B., Prasad, S. G., Menon, S., Tang, J., Parsons,
T., Ryan, D., Fisher, M., Williams, D., Dales, N. A., Patane, M.
A., and Pantoliano, M. W. (2004) ACE2 X-ray structures reveal
Biochemistry, Vol. 46, No. 30, 2007 8765
a large hinge-bending motion important for inhibitor binding and
catalysis, J. Biol. Chem. 279, 17996-18007.
54. Sturrock, E. D., Natesh, R., van Rooney, J. M., and Acharya, K.
R. (2004) Structure of angiotensin I-converting enzyme, Cell. Mol.
Life Sci. 61, 2677-2686.
55. Naqvi, N., Liu, K., Graham, R. M., and Husain, A. (2005)
Molecular basis of exopeptidase activity in the C-terminal domain
of human angiotensin I-converting enzyme: insights into the
origins of its exopeptidase activity, J. Biol. Chem. 280, 66696675.
56. Liu, X., Fernandez, M., Wouters, M. A., Heyberger, S., and Husain,
A. (2001) Arg(1098) is critical for the chloride dependence of
human angiotensin I-converting enzyme C-domain catalytic activity, J. Biol. Chem. 276, 33518-33525.
57. Nchinda, A. T., Chibale, K., Redelinghuys, P., and Sturrock, E.
D. (2006) Synthesis and molecular modeling of a lisinopriltryptophan analogue inhibitor of angiotensin I-converting enzyme,
Bioorg. Med. Chem. Lett. 16, 4616-4619.
58. Isaac, R. E., Michaud. A., Keen, J. N., Williams, T. A., Coate, S.
D., Wetsel, W. C., and Corvol, P. (1999) Hydrolysis by somatic
angiotensin-I converting enzyme of basic dipeptides from a
cholecystokinin/gastrin and a LH-RH peptide extended at the
C-terminus with gly-Arg/Lys-arg, but not from diarginyl insulin,
Eur. J. Biochem. 262, 569-574.
59. Wen, J. Y., Ledger, R., McLeod, B. J., Davies, N. M., Butt, A.
G., and Tucker, I. G. (2002) Enzymatic degradation of luteinizing
hormone releasing hormone (LHRH) by mucosal homogenates
from the intestine of the common brushtail possum (Trichosurus
Vulpecula), Life Sci. 71, 3019-3030.
60. Shimizu, T., Berisha, B., Schams, D., and Miyamoto, A. (2007)
Changes in the Messenger RNA expressions of the endothelin-1
and angiotensin systems in mature follicles of the superovulated
bovine ovary, J. Reprod. DeV., in press.
BI700253Q